Experimental investigation of the check valve behaviour when the flow is reversing

Expe imen al in es iga ion o he check al e beha iou when he flow is e e sing D. Him a,V.Hab ´ an, M. Hudec, and V. Pa l´ ık B no Uni e si y o Technology, Facul y o Mechanical Enginee ing, Ene gy Ins i u e, Vic o Kaplan Depa men o Fluid Enginee ing, Technick´ a 2, 616 69, B no, Czech Republic Abs ac . Check al e in a pipeline is supposed o p e en he e e se flow and o allow he flow in he posi i e di ec ion. The cons uc ion o check al es ollows hese equi emen s, bu he check al e mus no cause p essu e pulsa ions in ansien s. I means when he fluid is accele a ing o decele a ing. The a icle desc ibes an expe imen al in es iga ion o a swing check al e when he flow is changing i s di ec ion. The check al e was placed in an expe imen al ci cui , whe e he p essu e on he ups eam and downs eam side o he al e was measu ed and he cu en alue o flow a e was de e mined. The goal was o simula e condi ions in he eal sys em, whe e he check al e slam had been obse ed. 1 In oduc ion The check al e is an impo an pa o hyd aulic sys ems and allows flow in one di ec ion and p e en he e e se flow h ough he pump. The check al e can be also in- s alled in he sys em o limi he p essu e su ge induced by he pump ailu e [1]. The basic equi emen s on he check al e a e: – he low p essu e loss in he posi i e di ec ion, –no flow in he opposi e di ec ion, i means good sealing when he al e is closed. P oduce s, usually, gi e s a ic cha ac e is ics o check al es such as a p essu e d op dependence and opening de- pendence on he flow a e, he c acking p essu e ( he p es- su e when he check al e s a s opening) and he minimal flow a e when he check al e is ully open. The p oduce also can say whe he he al e is app o- p ia e o he pulsa ing flow, sludge wa e , ho izon al o e ical pipe and so on. These pa ame e s oge he wi h he eliabili y and main enance difficul y allow choosing he igh check al e o educe ope a ional cos s. Bu , he e is a p oblem wi h p edic ing he dynamic beha iou o he check al e in he pa icula sys em. I is, maybe, mo e impo an han p ope ies w i en abo e. When he fluid flow changes i s o ien a ion, he check al e is desi ed o close be o e he back flow eloci y becomes oo high o he wise he check al e disc slams and makes he high p essu e su ge, which is o en connec ed wi h he column sepa a ion pa icula ly on he ups eam side o he check al e. The lowe s a ic p essu e in he pipeline he g ea e column sepa a ion p obabili y. The slam can lead o se e e damages o he check al e and/o he whole sys em [2], [3], [4]. The check al e also can make p oblems wi h sel - exci ed p essu e pulsa ions: The oscilla ing con ol al e a he downs eam end o pipe causes oubles wi h check ae-mail: [email p o ec ed] al e slam a he ups eam end [5]. An example o he p es- su e su ge due o he check al e slam a e he pump s op- page is shown in he figu es 1 and 2. The slam also causes s ong mechanical ib a ions o pipeline wi h accele a ion abou 20 G in his case [4]. McElhaney made an ex ensi e analysis o check al e ailu es in he nuclea indus y, because hese ailu es a e well desc ibed, and hey ocused on co ela ion be ween he al e design and ailu e mode and ailu e dis ibu ion [6]. Time [ s ] 012345 -1 0 1 2 3 4 5 P essu e/Ope a ing p essu e [-] Downs eam side Ups eam side Fig. 1. P essu e su ge a he check al e due o slam [4] 3.5 3.7 3.9 4.1 4.3 4.5 Time [ s ] -1 0 1 2 3 4 5 P essu e/Ope a ing p essu e [-] Downs eam side Ups eam side Fig. 2. P essu e su ge a he check al e due o slam – de ail [4] DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 © The Au ho s, published by EDP Sciences. This is an open access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License 4.0 (h p://c ea i ecommons.o g/licenses/by/4.0/). In se e e pumping applica ions, almos all basic check al es will slam, and in ex emely mild applica ions, ha dly any check al es will slam [7]. Bu i is difficul o p edic p obabili y o he check al e slam be ween hese wo ex- emes. The calcula ion o he sys em decele a ion is impo an o say whe he he check al e will slam o no . The decel- e a ion is calcula ed om he o iginal o wa d flow eloc- i y and ime pe iod when he eloci y eaches ze o. I de- pends on many ac o s: complexi y o he sys em, ic ion, pump ine ia, s a ic head and so on. Thus, he decele a ion can be ha dly de e mined wi hou di ec measu emen o a leas a nume ical simula ion o he flow in he pipe line sys em. The maximal eloci y o he e e se flow depends on he check al e cons uc ion and on he sys em decele - a ion. The highe decele a ion he highe e e se eloci y can de elop be o e he check al e closes. Then, he p es- su e su ge ollows he Joukowski equa ion. The e e se eloci y dependence on he flow decele a- ion acco ding o [8] is shown in he figu e 3 and a depen- dence published in [9] is plo ed in he figu e 4. 0 1.0 1.5 2.0 Decele a ion [ m/s ] 0 0.2 0.4 0.6 0.8 1 . 0 Re e se eloci y [m/s] 0.5 2 Swing check Bi old Til ed disc Fig. 3. Re e sal eloci y o diffe en cons uc ions o he check al e by [8] 01015 Decele a ion [ m/s ] 0 0.1 0.2 0.3 0.4 0 .5 Re e se eloci y [m/s] 5 2 Swing check Til ed disc check Dual disc check Ball check Swing lex check Silen check Nozzle check Fig. 4. Re e sal eloci y o diffe en cons uc ions o he check al e by [9] One can see ha esul s (e. g. o he il ed disc check al e) a e qui e diffe en so i is ob ious ha he e e - sal eloci y does no depends only on he decele a ion and ype o he check al e, bu , p obably, also on he specific cons uc ion, size, mass, placemen ,. . . The sys em decele - a ion gi es jus a basic hin whe he he check al e ends o slam o no . Table 1. S and specifica ions En y Value Uni Pipe diame e 0.1 m Maximal flow a e 31 l s−1 Maximal flow eloci y 3.95 m s−1 Leng h o discha ge pipe 17.3 m S a ic head 7.5 m Volume o bo om ank 1 m3 Volume o op ank 0.8 m3 Pipe ma e ial s eel Maximal pump speed 1450 pm 2 Expe imen An expe imen al s and has been buil and se es o he examina ion o he check al e beha iou , when he flow o ien a ion is changing. The expe imen is a pa o co- ope a ion wi h MSA company, a p oduce o al es, and he goal is o design a check al e, which does no slam in ex eme ope a ing condi ions. MSA company p o ided a swing check al e DN 100, which ends o slam and i is he fi s check al e o a ow, which is going o be es ed. The s and consis s o he pump, which collec s wa e om he bo om ank. The es ed check al e is placed in he ho izon al pa o he discha ge pipe 2.3 m abo e he wa e le el in he bo om ank. The discha ge pipe ends in he op ank. The wa e om op ank e u ns back h ough he e u ning pipe and he o e spill pipe (see figu e 5 and able 1). P essu e senso s ( ange 0 – 1 MPa, unce ain y 0.25% o he ange, sampling equency 1 kHz) a e placed 0.3 m, 1.4 m and 2.5 m be o e he check al e and 0.41 m, 1.4 m and 2.38 m behind he check al e in he s eady flow di- ec ion. The flow a e is measu ed wi h he elec omagne ic flowme e (0 – 80 l/s, unce ain y 0.5% o measu ed alue). One accele ome e (-50 G – 50 G, unce ain y 1% o mea- su ed alue, sampling equency 1 kHz) is placed on he check al e body and ano he one is placed on he pump. Accele ome e s should help o iden i y an exac ime o some e en s a e pump disconnec ion. The pump speed is measu ed wi h he lase and one ma k on he mo o sha . The pump speed is con olled wi h a equency con- e e o con ol he flow a e and flow decele a ion. Un- o una ely, he e was a p oblem wi h he con e e du ing he expe imen s and i was no possible o con ol he de- cele a ion a e. The check al e design does no allow measu ing po- si ion o he disc, bu check al es, which a e going o ol- low, ha e windows o see he exac posi ion o he disc and allow PIV measu emen o he flow field in he check al e body. Th ee p essu e senso s on bo h sides o he al e allow calcula ing he wa e speed and flow a e du ing decele a- ion by ime-p essu e me hod. Using he flowme e is no possible due o i s g ea in eg a ion cons an so he flowme- e can measu e jus he s eady flow a e. The expe imen p ocedu e is ollowing: 1. S a o pump and un wi h he specific speed o ha e a desi ed flow a e ( om 10 l/s o30l/s). I was no pos- sible o keep he flow o ime longe han 30 s, because he o e spill ae a ed wa e . The ai dec eases he wa e DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 2 p Q p 46 O e spi ll Discha ge pipe Re u ning pipe O e spill pipe 7.5 m 7.5 m DN 100 DN 100 DN 100 p 1p 2p 3p 5 2.3 m Fig. 5. Expe imen al s and speed, which di ec ly influences he heigh o p essu e peaks acco ding o Joukowski equa ion, and changes he sys em esponse [10], [11]. 2. As soon as he s eady flow was eached he pump was disconnec ed. Since he pipe line is qui e sho , i was possible o u n off he pump a e fi e seconds a e he s a . 3. The measu emen was eco ded om he s a o pump un il all ansien s a e he pump s op we e finished. Then, i is possible o e alua e: c acking p essu e, flow decele a ion, maximal back flow eloci y, p essu e su ge in on o and behind he check al e, wa e speed and accele a ion connec ed wi h mechanical ib a ion on he check al e body. 3 Resul s The pump was s opped om he ou diffe en s eady flow a es: 10 l/s, 15 l/s, 20 l/s, 27 l/s and 30 l/s. An example o esul s o s op om 15 l/s is shown in he figu es 6 o 12. The ins an pump speed is e alua ed om scanning he spo on he mo o sha . I allows iden i ying he ime o pump disconnec ion om he g id and he ime, when he sha defini ely s opped. These poin s a e, espec i ely, la- belled wi h numbe s 1 and 6 in he figu es. Since he e was only one ma k on he o o sha , he e is no p oblem o coun numbe o e olu ions and calcu- la e he exac speed ( he e is no any demand on he exac posi ion o he ma k), bu (because we a e alking abou an uns eady o a ion) he slowe sha o a es he less accu a e e alua ed speed is. Tha is he eason why he las e alu- a ed speed o he sha is 85 pm. The las ma k was ead a he ime 3.2 s, he sha s opped a e ha . P essu e ups eam o he check al e (figu e 7) ex- hibi s he pump disconnec ion (poin 1), when he p essu e s a ed dec easing. Time, when he flow changes o ien a- ion and s a s flowing backwa d, is ma ked wi h poin 2. This e y weak peak can be easily o e looked. Poin 3 labels ime ins an , when he check al e s a s closing and poin 4 is he momen , when he disc hi s he 012345 0 200 400 600 800 1000 Time [ s ] S peed [ pm] 1 2 4 6 3 5 Fig. 6. Pump decele a ion om s eady flow a e 15 l/s sea . (The column sepa a ion occu ed he e in cases wi h g ea e ini ial flow a e). Then, he highes peak eme ges and p essu e pulsa ions a e damped, bu he pump sha s ill o a es. The sha s op (poin 6) caused amplifica ion o he ollowing pulsa ions abou 100%, bu he absolu e alue was s ill low, hus i did no cause any significan e en . These amplified pulsa ions we e damped wi hin 1 second. 0 1 2 3 4 5 0 50 100 150 200 2 5 0 Time [ s ] P essu e [kP a ] 1 3 2 6 4 5 Fig. 7. P essu e in on o he check al e du ing he ansien DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 3 The p essu e downs eam o he check al e (figu e 8) allows iden i ying he ime when he pump was discon- nec ed (poin 1). Poin s 3 and 4 ma k momen when he disc s a ed and finished closing. The ollowing p essu e d op goes o he alue, which is lowe han one would expec (poin 5). The same poin in he figu e 7 shows ha he p essu e peak is lowe han i should be, when he p e ious and subsequen peaks a e conside ed. This is he ime ins an when he disc eopens (p essu e a he ups eam side o he disc is g ea e han a he downs eam side) and closes again. The figu e 9 0 1 2 3 4 5 80 100 120 140 160 180 200 220 2 4 0 Time [ s ] P essu e [kP a ] 1 2 4 6 35 Fig. 8. P essu e behind he check al e du ing he ansien shows he p essu e diffe ence on he check al e. A posi- i e alue co esponds o he si ua ion, when he p essu e downs eam o he al e is g ea e han ups eam o he al e, so he check al e is closed. Nega i e alue a he poin 5 suppo s he s a emen ha he disc eopens a ha momen . 0 1 2 3 4 5 50 0 50 100 150 200 Time [ s ] P essu e di e ence [kPa] 1 2 46 3 5 Fig. 9. P essu e diffe ence on he check al e du ing he ansien The flow a e was e alua ed om he p essu e eco ds p4and p6by he ime-p essu e me hod: Q( +Δ )=Δ S Lρp4−p5−RQ2( )+Q( ),(1) whe e he esis ance Ris no a unc ion o he ime, bu depends on he ini ial flow a e (2), which is subjec o he nume ical op imiza ion. R=p4(0)−p5(0) Q2(0).(2) The dependence o he eloci y on he ime (figu e 10) is compu ed om (1) by an app op ia e nume ical me hod, when he flow a e a he las ime ins an equals ze o. The solu ion also includes he ini ial flow a e and esis ance. Then, he s a o decele a ion (poin 1) can be easily ound as well as he ime, when he flow changed i s o ien a ion (poin 2). The maximal backwa d eloci y (poin 4) is he las in o ma ion ob ained om his g aph. 0 1 2 3 4 5 0 0.5 1 1.5 2 Time [ s ] Flow eloci y [m/s] 1 24 6 3 5 Fig. 10. Flow eloci y du ing he ansien The measu emen o he accele a ion on he pump body also gi es some in o ma ion abou e en s in he pipeline (see figu e 11). The pump disconnec ion is no isible he e ( he e is no any change o he signal a poin 1), because all mo ing pa s kep hei mo ing. The fi s change is ob i- ous when he speed dec eases abou 50%. The accele a ion ampli ude becomes no iceably lowe 0.3 s a e he pump disconnec ion. Closu e o he check al e disc is well isible (poin 4) as well as i s second closu e a e eopening (poin 5). The pump sha s op (poin 6) is be e ecognizable han pump disconnec ion. 012345 0.4 0.2 0 0.2 0.4 0 . 6 Time [ s ] Accele a ion [ G ] 1 2 4 6 3 5 Fig. 11. Measu emen o he accele a ion on he pump body The signal om he accele ome e on he check al e body (figu e 12) con ains s ong noise so e en he iden i- fica ion o he exac ime o disc closu e (poin 4) is diffi- cul . The e is a ques ion whe he he noise came om me- chanical easons (e. g. pipeline ib a ions) o om signal in e e ence. The signal in e e ence is he mos p obable, because mechanical ib a ions would be isible also on he accele a ion signal om pump body. I is in e es ing ha he s onges peak co esponds o he second p essu e peak downs eam o he check al e (see figu e 8), bu , a he same momen , he p essu e up- s eam o he check al e is e y low. I is he ime, when he disc closes o he second ime. DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 4 012345 0.4 0.2 0 0.2 0.4 0 . 6 Time [ s ] Accele a ion [G] 12 4 6 3 5 Fig. 12. Measu emen o he accele a ion on he check al e body Table 2. Compu ed decele a ion and he maximal e e se eloci y Ini ial flow eloci y Decele a ion Re e se eloci y (m s−1)(ms −2)(ms −1) 1.3 2.4 0.15 1.9 2.9 0.22 2.5 3.5 0.26 3.4 4.1 0.34 3.8 4.7 0.43 The sys em decele a ion can be easily compu ed om he figu e 10 using poin s 1 and 2. Maximal e e se flow eloci y can be ead a he same figu e. These da a a e lis ed in he able 2 o diffe en ini ial flow eloci ies. Figu e 13 shows measu ed da a compa ed wi h da a p o ided by [9]. The ob ained cha ac e is ics is ound be- ween cha ac e is ics o swing check al e and swing flex check al e. 03 5 Decele a ion [ m/s ] 0 0.1 0.2 0.3 0.4 0 .5 Re e se eloci y [m/s] 1 2 42 Swing check Swing lex check Measu ed swing check Fig. 13. Maximal e e se eloci y dependence on he sys em de- cele a ion 4 Discussion Following lis summa izes impo an e en s in he sys em ( he numbe ing co esponds o he numbe s in he figu es): 1. Pump disconnec ion – he pump s a s decele a ing, which isible on he di ec measu emen o pump e - olu ions and p essu e, because he p essu e change is p opo ional o he second powe o speed change. This poin is also well ecognizable on he compu ed flow eloci y, bu accele a ion on he pump body does no gi e any clue. 2. Change o he flow o ien a ion – can be iden ified om he compu ed flow eloci y. A low peak can be ound also in he signal om he p essu e ansduce ups eam o he check al e. 3. S a o he disc closu e – is easily isible on he p es- su e eco d a bo h sides (up- and downs eam) o he check al e. 4. End o he disc closu e – is easily isible on he p es- su e eco d and also on he compu ed flow a e. I co - esponds o he maximal e e se flow eloci y. This poin could be also iden ified on he accele a ion o he pump body, bu , su p isingly, i was e y difficul o find exac poin also on he accele a ion o he check al e body. This eco d is e y noisy. 5. Disc eopening and closing – can be ound on he p essu e signals and i is isible also on he accele - a ion o he pump body. 6. Pump sha s op – is defini ely ecognizable wi h scan- ning he sha , bu i s speed be o e he s op is ques ion- able, because only one ma k on he sha has been used. I is also possible o iden i y his poin on he signal om he accele ome e on he pump body and on he p essu e a downs eam side o he check al e. Resul s show ha he highe ini ial flow he mo e in- ense is decele a ion a e pump disconnec ion. I can be caused by highe ic ion loss. I is also appa en ha he highe decele a ion he highe e e se eloci y de elops be o e he check al e closing, because he disc shu s wi h longe delay. The check al e migh no be ully open o lowe flow a es, so closes as e . Ano he explana ion could be ha highe decele a ion causes swi ls in he check al e space. These swi ls migh suppo he disc and cause de- layed closing. 5 Conclusion The pape is ocused on he measu emen o dynamic cha - ac e is ic o he swing check al e. An expe imen al ci - cui has been buil and he swing check al e beha iou was obse ed du ing ansien e en s occu ing a e pump disconnec ion om he elec ic ne wo k. In es iga ed check al e is fi s o a ow o check al es wi h diffe en cons uc ions, which a e going o be es ed. This es was supposed o show whe he we a e able o measu e dynamic cha ac e is ic o he check al e and i i is possible o desc ibe ansien e en occu ing in he sys- em. The ollowing expe imen s a e going o include also eco ding o he check al e disc mo ion. Acknowledgemen This wo k has been suppo ed by Technology Agency o he Czech Republic unde he p ojec Inno a i e esea ch o check al es o ex eme ope a ing condi ions in ene - ge ics TH01011352. Nomencla u e L(m) pipe leng h be ween p essu e ans- duce s, DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 5 p(Pa) p essu e, see figu e 5 Q(m3s−1) flow a e, R(Pa s2m−6) esis ance S(m2) pipe c oss-sec ion, (s) ime, Δ (s) ime s ep, ρ(kg m−3) densi y. Re e ences 1. B. W. Ka ney, A. R. Simpson, Jou nal o Hyd aulic Re- sea ch 45 547 – 554 (2007) 2. C. Chiu, Ame ican Nuclea Socie y 54, 289 – 291 (1987) 3. J. R. T a is, M. D. To ey, Ame ican Socie y o Me- chanical Enginee s win e annual mee ing, 1 – 9 (Miami 1985) 4. D. Him , V. Hab´ an, P. Dokoupil EPJ Web o Con e - ences 114 (2016) 02038 5. D. Him , V. Hab´ an, Applied Mechanics and Ma e ials 630, 375 – 382 (2014) 6. K. L. McElhaney, Nuclea Enginee ing and Design 197, 169 – 182 (2000) 7. J. V. Ballun Jou nal o AWWA 99, 3 (2007) 8. W. Rahmeye Nuclea Indus y Check Val e G oup, 1996 Win e Mee ing,1–10(S . Pe e sbu g1996) 9. Val-Ma ic Val e and Manu ac u ing Co p. Dynamic Cha ac e is ics o Check Val es (2003) 10. J. Jablonsk´ a, M. Kozubko ´ a, EPJ Web o Con e ences 114 (2016) 02049 11. M. Vaˇ sina, L. H uˇ z´ ık, A. Bu eˇ cek, EPJ Web o Con- e ences 114 (2015) 02131 DOI: 10.1051/ ,02036 (2017) 714302036 143 EPJ Web o Con e ences epjcon /201 EFM 2016 6